欢迎访问林业科学,今天是

林业科学 ›› 2019, Vol. 55 ›› Issue (11): 52-62.doi: 10.11707/j.1001-7488.20191107

• 论文与研究报告 • 上一篇    下一篇

基于模拟退火算法逆转搜索的森林空间经营规划

孙云霞1,刘兆刚1,2,董灵波1,2,*   

  1. 1. 东北林业大学林学院 森林生态系统可持续经营教育部重点实验室 哈尔滨 150040
    2. 黑龙江头雁创新团队计划——森林资源高效培育技术研发团队 哈尔滨 150040
  • 收稿日期:2017-12-31 出版日期:2019-11-25 发布日期:2019-12-21
  • 通讯作者: 董灵波
  • 基金资助:
    国家自然科学基金项目(31700562);中央高校基本科研业务费专项资金(2572019BA07)

Spatial Forest Management Planning Based on Reversion Search Technique of Simulated Annealing Algorithm

Yunxia Sun1,Zhaogang Liu1,2,Lingbo Dong1,2,*   

  1. 1. Key Laboratory of Sustainable Forest Ecosystem Management, Ministry of Education School of Forestry, Northeast Forestry University Harbin 150040
    2. Heilongjiang Touyan Innovation Team Program:Technology Development Team for High-Efficient Silviculture of Forest Resources Harbin 150040
  • Received:2017-12-31 Online:2019-11-25 Published:2019-12-21
  • Contact: Lingbo Dong
  • Supported by:
    国家自然科学基金项目(31700562);中央高校基本科研业务费专项资金(2572019BA07)

摘要:

目的: 研究逆转搜索策略对模拟退火算法性能的提升作用,为复杂森林经营规划问题的优化求解提供技术支撑。方法: 以大兴安岭塔河林业局盘古林场森林空间收获安排问题为例,以模拟退火算法1-邻域和2-邻域技术为基础构建逆转搜索过程,并将其应用于具体的森林经营规划实践。规划模型以10年规划期内的木材均衡收获为目标函数,根据经营措施选用不同空间约束形式,对抚育措施采用面积限制模型,对皆伐措施采用单位限制模型,2种模型均满足3年的绿量约束期。结果: 对于最小化森林经营规划问题而言,逆转搜索中不同邻域间的交互次数对规划结果影响不显著,但其平均目标函数值均显著低于传统模拟退火算法1-邻域(P < 0.01)和2-邻域(P < 0.01),而算法优化时间分别比1-邻域和2-邻域增加约5和2倍,逆转搜索策略具有显著的优越性能和广泛的应用前景;最优森林经营方案表明,整个规划期(10年)内预计可收获木材约5.00×105 m3,其中抚育出材量3.12×105 m3,抚育林分面积占林场总面积的10.94%,皆伐出材量1.88×105 m3,皆伐林分面积占林场总面积的1.02%,该方案能够满足森林可持续经营的目标。结论: 逆转搜索是一种高效、稳定的优化求解技术,能够满足复杂森林经营规划问题的需要,可为森林经营方案的编制提供技术支撑。

关键词: 模拟退火算法, 逆转搜索, 经营规划, 邻接约束, 绿量约束

Abstract:

Objective: The effects of reversion search strategy on improving the performance of simulated annealing algorithm were quantificationally evaluated,which could provide some new insights into solving the complex forest planning problems. Method: The reversion search process were estimated based on the 1-opt and 2-opt moves of simulated annealing,and then were applied into a large and real forest planning in Pangu forest farm in the northeastern part of China. The objective function of the planning problem focused on the needs for an even flow of harvest volume. A special strategy was employed to deal with the spatial and temporal distribution of forest management activities,in which the forest tending prescriptions complied with the unit restriction model of adjacency constraints,however the clear-cutting prescription complied with the area restriction model of adjacency constraints. In addition,both adjacency models should meet 3-yrs green-up constraints. Result: The results indicated that the number of reversions between 1-opt and 2-opt moves usually had no significant effects on the planning results. For a minimization planning problem,the mean objective function values of reversion search decreased significantly when compared that with 1-opt (P < 0.01) and 2-opt (P < 0.01) moves,respectively,however,the mean computation time of reversion search was only as large as five and two times than that of 1-opt and 2-opt moves,respectively. The optimal solution indicated that the amount of assigned harvest timber during an entire planning horizon were approximately 5.00×105 m3,in which the levels of harvest timber using different forest tending prescriptions were 3.12×105 m3,and the levels of harvest timber using clear-cutting prescription were 1.88×105 m3,the assigned harvest area of forest tending and clear-cutting only accounted for approximately 10.94% and 1.02% of the total forest area,respectively. Therefore,the optimal solution from this study could meet the requirements of sustainable forest management. Conclusion: Reversion search is one of the most efficient and stable strategy to improve the search ability of heuristics,which could adapt to the complex forest management planning,and then provide some technical supports for making approximate forest management plans in future.

Key words: simulated annealing, reversion search, management planning, adjacency constraints, green-up constraints

中图分类号: